Relapse after therapy remains a critical problem for most types of cancer in humans, even when cancer- specific therapy is initially effective. Understanding the basis for cancer resistance and recurrence is a critical pre-requisite for developing more effective treatments. Recent data indicate that in some cancers, including leukemia, colon cancer, and certain brain tumors, a minority of cells in the initial tumor possesses unique properties reminiscent of normal tissue stem cells: a capacity for extensive self- renewal and multilineage differentiation potential. Relative drug resistance in these cells may be especially relevant for cancer recurrence. Thus, the precise relationship between these tumor-initiating cells and normal tissue-specific stem or progenitor cells is extremely important to understand in depth, as key differences might be exploited for therapeutic benefit. In this proposal we outline an approach to identify critical genome-wide similarities and differences between normal and cancer stem cells isolated from normal intestine/colorectal adenocarcinoma, neural stem cells/glioblastoma multiforme (GBM) and hematopoietic stem cells/acute myelogenous leukemia (AML). Using well-validated reagents and experimental approaches, we will isolate highly purified and functionally characterized cancer stem cells from primary human tumors and well-defined mouse models of cancer. We will interrogate gene expression, histone and DNA methylation, and microRNA expression for comparison to normal tissue stem cells, which we will isolate in parallel for these studies. Detailed comparative data will be deposited in a well-annotated central database and analyzed in a manner that allows comparison of normal and malignant stem cells from multiple different tissues. The collective data will provide an unprecedented and comprehensive picture of normal and cancer stem cells, highlighting key differences that may be exploited therapeutically. The dataset and concomitant analysis, served on a powerful computational platform and web interface, will serve as an invaluable resource for the stem cell and cancer research communities. To complement these efforts, we will also develop sophisticated mouse reporter models of four signaling pathways that play seminal roles in cancer stem cell biology: Wnt/ss-catenin, Sonic hedgehog, Notch, and PIK3CA/FoxO. As our genomic approaches identify new stem cell pathways, in year 2 we will be positioned to develop up to 3 additional mouse reporter models that will both validate our genomic findings and provide critical resources for future study of these and other tumor types. In sum, the studies proposed in this highly collaborative project will provide vital information to aid in developing therapies that might target cancer stem cells while sparing their normal counterparts. The nature, scope and future scientific and economic impact of this work are ideally suited to the goals of the "Grand Opportunities" funding mechanism, and indeed, in-depth comparison of normal and tumor stem cells can only be accomplished under the aegis of such a program. PUBLIC HEALTH RELEVANCE: Many cancers including brain tumors, colon cancer and leukemia contain a population of cancer stem cells that are similar to normal stem cells in some respects, including an inherent ability to propagate the tumor;these "cancer stem cells" are likely resistant to current therapies. This proposal will identify the common and shared feature of cancer stem cells and normal stem cells derived from 3 tissues: blood, colon and brain. The results of these studies will inform development of new treatment strategies that target cancer but spare normal stem cells, leading to the development of more effective and less toxic cancer treatments.